Safety-critical autonomous driving requires reliable decision-making in a highly interactive, safety-critical urban driving scenario. Although learning-based methods are attractive for such problems, unconstrained trial-and-error exploration can readily produce unsafe actions during training and lead to unreliable closed-loop behavior. This paper considers an unprotected left-turn task at a signal-free intersection and proposes a safety-guided reinforcement learning (RL) framework in which a Control-Barrier-and-Lyapunov-Function-based Quadratic Program (CBLF-QP) serves as a guidance mechanism for policy learning rather than as a hard safety shield. The framework is built on an affine kinematic vehicle model, and higher-order control barrier function constraints are used to encode geometric safety requirements, including collision avoidance and lane keeping. At each state, the CBLF-QP computes a minimum-intervention safe reference action together with a barrier-violation indicator. The reference action is incorporated into policy learning through a mean-squared imitation loss, progressively reshaping the policy distribution without the discontinuities caused by hard action replacement. A primal-dual adaptive weighting mechanism further modulates the guidance strength according to the violation level, balancing task performance and safety regularization during training. Simulations under different traffic-density levels and real-vehicle experiments with virtual surrounding traffic demonstrate improved training stability, reduced collision-related safety cost, and practical feasibility while maintaining competitive efficiency.

Strong switching nonlinearities in electromechanical brake (EMB) systems pose great challenges for clamping force control. This article proposes an integrated linear closed-loop control framework for a nonlinear switching EMB system. First, the linear modeling is enabled by the physics-informed Koopman operator for friction torque and a linear separation strategy for nonlinear clamping force. Then, the real-time clamping force estimation is fulfilled by designing a Kalman filter with unknown input that can deal with arbitrary brake switching. Finally, a triple-loop cascaded PID control is adopted for accurate clamping force control. The closed-loop stability is analyzed using linear control theory thanks to the linear framework. Production-level vehicle equipped with EMB systems is employed for thorough validation. Under moderate braking conditions, the relative root-mean-square errors of friction torque modeling, clamping force estimation, and tracking control are less than 8.4%, 4.6%, and 6.5%, respectively. Under emergency braking conditions, the RRMSE of clamping force estimation is 8.6%, and the settling time of force control is less than 10 ms. These results confirm the effectiveness and real-time capability of the proposed linear control framework for nonlinear EMB system.

This study addresses the challenge of recognizing anomalous intent in surrounding vehicles for autonomous driving, proposing the ADID+ framework. The framework employs dynamic frequency band division and dual-stream attention mechanisms to integrate vehicle-road-cloud collaborative data, achieving high-precision, low-latency anomaly detection. Evaluated on 12,586 multi-modal annotated samples covering six anomaly types (such as malicious cut-ins and illegal overtaking), ADID+ demonstrated an F1-score improvement of 6.5%-18.6% over mainstream methods. It also reduced the average detection delay to 132ms, thereby meeting active safety system requirements. This approach overcomes the bottlenecks of traditional time-domain models through frequency-domain feature enhancement, revealing distinct spectral representation patterns for abnormal behaviors like malicious cut-ins (5.2Hz) and illegal overtaking (3.5/4.2Hz). Even in scenarios with missing multi-modal data or extreme parameters, the framework maintains a detection accuracy above 72% and a false alarm rate below 3.7%. Limitations of this study include not accounting for communication latency jitter and online frequency band optimization. Future work will focus on developing dynamic latency compensation mechanisms and lightweight operators to enhance robustness under non-ideal conditions.

Electrification,intelligence,and multifunctionality can be promoted in agricultural machinery.In this study,a distributed electric tracked chassis was specifically developed for complex unstructured agricultural terrain.A modular chassis also consisted of the power,walking,electrical control,and intelligent perception system.Among them,the power system was driven by a 72 V lithium battery pack.The power was then provided for two 3 kW permanent magnet synchronous servo motors,each of which was integrated with a planetary gearbox to output the high torque suitable for track drive.The walking system was integrated with the rubber tracks,drive wheels,guide wheels,an independent suspension wheel group,and a tensioning device,thus balancing lightweight with high passability.The electrical control system consisted of the vehicle control unit,motor control units,battery system,and CAN communication network.The torque and energy distribution were responsible for the real-time data interaction with the subsystems.The intelligent perception system was integrated with a GPS/IMU navigation device,PTZ camera,and ultrasonic radar using the intelligent driving domain controller.Precise positioning,environmental perception,and path planning were then achieved in complex agricultural working environments.In the electrical and electronic architecture,the"high-voltage drive-low-voltage control"layered power supply and multi-bus communication topology were adopted to adjust the rapid dynamics and high reliability of the power system.The high-voltage system was powered by the 72 V battery,with the energy distribution and protection using the high-voltage distribution box.The low-voltage system consisted of the 12 and 24 V networks,providing power to the controllers,sensors,communication devices,and braking units.The multi-bus collaborative communication topology included multiple CAN buses and an RS485 bus,the real-time data exchange between the vehicle and motor control units,environmental perception and intelligent control data exchange,as well as the program burning and parameter calibration.The coordinated multi-bus system was obtained through efficient and stable data transmission between control units,providing reliable communication support for the intelligent operation.According to the centralized domain controller and standardized V-model,the application layer software was developed for the chassis domain controller.A model-based design was integrated with the multi-level control functions,signal diagnostics,and functional safety mechanisms.The application layer software was verified the automatically generated code,fully meeting the standard requirements of ISO 26262 functional safety,MAAB modeling,MISRA C 2023 code generation,the software compliance and functional safety.Multi-body dynamic rigid-flexible simulations and prototype testing show that the distributed driven electric tracked chassis was also met the requirements of the power,passability,load capacity,climbing,obstacle-crossing and steering performance.In the maximum speed test,the chassis was operated stably at 2.11 m/s;In the load performance test,the chassis ran smoothly with a 1 500 kg load,with the even track grounding and no slippage or yaw;In the 25° slope climbing test,the chassis demonstrated the stable climbing;In the obstacle-crossing test,the chassis successfully passed the obstacles up to 0.25 m high,with the minimal vertical acceleration and no severe shock or subsidence;In the steering performance test,the chassis exhibited the better trajectory consistency during in-place and small-radius turns,with the maximum deviation controlled within±2％,and the minimum turning radius 12％smaller than the design value,indicating the excellent steering flexibility and structural stability.Overall,a systematic technical pathway can be obtained for the electronic,electrical,and controller software architecture of the distributed electric-driven crawler chassis for complex agricultural working environments.The finding can also provide the technical and theoretical support for the high-performance and highly adaptable intelligent agricultural machinery.

Distributed drive electric vehicles (DDEVs) represent a promising chassis configuration for electric vehicles, offering superior handling performance due to their over-actuated characteristics. However, this over-actuation also intensifies the inherent conflict between maneuverability and stability. To address this challenge, a novel concept of the maneuverable safety region (MSR) is proposed, which quantifies the safe executable boundaries of traction force and direct yaw moment. Based on the MSR, five vehicle handling modes corresponding to different sub-regions are defined to qualitatively analyze the degree of conflict between maneuverability and stability, with linear inequalities employed to construct the dynamic boundaries of each mode. Furthermore, an innovative multi-mode torque vectoring control strategy (MTVS) is developed to satisfy the control requirements of different modes, considering tire adhesion utilization, energy consumption, and tire slip energy dissipation. Finally, comprehensive algorithm tests demonstrate that the MTVS effectively achieves a trade-off between maneuverability and stability, while simultaneously ensuring vehicle safety and controllability at handling limits.

Freeway on-ramp merging represents a high risk conflict zone in which control transition failures in conditionally automated driving (AD) can contribute to severe collisions. Such events are often linked to a mismatch between rapidly shrinking safety margins and driver cognitive readiness at takeover. Conventional takeover warning strategies that rely on static kinematic thresholds may not account for the driver dynamic pre-crash state including trust, risk perception, and driving style. This limitation can increase the likelihood of missed alarms and nuisance alarms and can reduce the effectiveness of warnings for accident prevention. To address this safety gap, this study proposes a personalized accident prevention framework that identifies high risk states and triggers predictive intervention before physical conflict becomes critical. The method integrates a Kalman filter based dynamic trust estimator with a learned risk perception module to construct a driving style sensitive joint trigger mechanism. The mechanism issues warnings when the driver trust state deviates from a personalized trust interval that represents a safe operating envelope under prevailing risk. Two independent simulator experiments were conducted to calibrate style specific trust intervals and to validate the proposed strategy in complex merging scenarios. Comparative results show improved safety outcomes relative to a fixed threshold baseline. The proposed strategy reduced mean collision rate from 40.46% to 17.05% and increased correct takeover rate from 51.23% to 64.22%. These findings indicate that integrating dynamic human state estimation and driving style heterogeneity into takeover warning logic can support prevention of takeover related accidents and can enhance safety resilience in automated merging operations.

3D object detection is a core task in autonomous driving perception systems, where LiDAR provides precise geometric structure information and cameras contribute rich semantic features. However, existing multi-modal detection frameworks often struggle to balance accuracy, robustness, and real-time performance in scenarios with sparse point clouds, insufficient semantics, and high computational overhead. To address this challenge, this paper proposes a novel multi-modal 3D object detection framework based on multi-stage cross-modal fusion. Specifically, the LiDAR branch integrates lightweight pillar encoding with BEV transformation and leverages a region proposal network (RPN) to efficiently extract spatial ROIs while preserving detailed geometric information. The image branch extends the YOLO network to 3D detection, enabling unified 3D bounding box prediction within a common coordinate system. In the fusion stage, an ROI-Attention module is designed, where the Translation-Aware Module (TAM) achieves cross-modal geometric alignment and the Region-Aware Module (RAM) explicitly mines complementary semantic and geometric features. The resulting unified feature volume is then fed into a lightweight detection head for classification and 3D bounding box regression. We validate the proposed method on both the KITTI dataset, the nuScenes dataset and real-vehicle experiments. Results demonstrate that the framework significantly improves detection accuracy and robustness while maintaining real-time performance, offering clear advantages in detecting small, occluded, and distant objects, and exhibiting strong adaptability in complex autonomous driving scenarios.

Accurate joint estimation of vehicle lateral states and actuator faults remains challenging in the presence of time-varying dynamics, actuator faults, and external input disturbances. This paper proposes a fuzzy proportional-integral (PI) adaptive robust observer for joint state-fault estimation under such conditions. A variable-parameter Takagi-Sugeno fuzzy model is first developed to describe the time-varying lateral dynamics induced by variations in longitudinal velocity and tire cornering stiffness. An augmented estimation structure is then constructed to enable unified reconstruction of vehicle lateral states and actuator faults within a single observer. To improve fault-estimation performance, a PI-type adaptive law is introduced to enhance transient response and reduce steady-state bias. In addition, an observer synthesis method combining an adjustable H infinity performance criterion, an auxiliary optimization index, and region-constrained pole placement is developed to improve disturbance attenuation and convergence performance while alleviating design conservatism. Hardware-in-the-loop and real-vehicle experiments are conducted under actuator fault and external input disturbance conditions. Experimental results show that the proposed observer achieves accurate and robust joint estimation of vehicle lateral states and actuator faults under time-varying operating conditions.

This paper proposes a Transformer-Enhanced Multi-Agent Reinforcement Learning (TE-MARL) framework for electric logistics fleet routing under time-varying traffic and coupled power-grid loads, with the goal of minimizing overall operating cost. First, we develop a linear congestion model with a 30-min smooth transition to approximate the continuous dynamics of traffic flows. In parallel, we establish an energy consumption-charging coupling model that captures the effects of congestion, speed, specific energy consumption, and peak-valley grid-load fluctuations on charging efficiency. Next, we design a three-module Transformer-based policy network that incorporates traffic-aware attention to strengthen the representation of congestion features. The framework employs centralized training with decentralized execution (CTDE) and Proximal Policy Optimization (PPO) to enhance generalization and stability in highly dynamic environments. Beyond conventional objectives such as minimizing travel time and energy consumption, we introduce a Traffic Adaptability Cost metric to quantify the robustness of routing solutions to traffic fluctuations. Extensive experiments under high station density and multi-vehicle coordination compare TE-MARL with state-of-the-art learning-based and heuristic baselines. The results indicate that, on a representative C100-S20-V8 instance, TE-MARL reduces total travel time and total energy consumption by 20.4% and 17.6% relative to a MARL baseline, and by 14.9% and 10.9% relative to a hybrid Q-DH method, while achieving 100% feasibility and higher traffic adaptability. These gains substantially mitigate the risks of peak-hour time-window violations and excessive energy use.

Driving on slippery extreme condition poses an huge challenge for automated electric vehicles (AEVs) due to the significant degradation of tire–road friction and the strong coupling between longitudinal and lateral vehicle dynamics. The performance of traditional model predictive control (MPC) schemes relies heavily on accurate vehicle dynamics models and carefully tuned cost functions, both of which are difficult to achieve under slippery curved road conditions.To address these challenges, a hybrid vehicle motion model is first constructed by integrating a physics-based vehicle dynamics model with a 1-Lipschitz residual neural network, which improves prediction accuracy and adaptability under uncertain road conditions. Based on the hybrid model, an MPPI controller is developed in which actor–critic networks are employed to adaptively tune the stage cost weight factors, while a temporal-difference (TD) learning algorithm is used to learn the terminal cost from selected MPPI rollout trajectories. In this manner, short-horizon control performance and long-term cost optimization are jointly addressed within a unified MPPI framework.Compared with model-free reinforcement learning approaches, the proposed controller preserves high physical interpretability and enhanced safety awareness by embedding learning mechanisms into predictive control rollouts, rather than directly applying learned actions to the vehicle. Extensive simulation results demonstrate the effectiveness and superiority of the proposed RL-HMPPI controller under slippery extreme road conditions.

Active collision avoidance is a crucial feature for autonomous driving of industrial vehicles, which requires the vehicles to navigate around obstacles while maintaining its multidimensional stability within a limited distance. To satisfy the requirements of safety and transportation efficiency simultaneously, this paper proposes a novel hierarchical control scheme based on a redefined collision avoidance distance and the modified stable handling envelope. First, in the upper decision-making level, the safety distance and minimum collision avoidance distance are determined subject to the multi-dimensional stability of the industrial vehicle. Then two collision avoidance cases are identified. In the first case, the reference acceleration is generated based on the modified stable handling envelope and the boundaries are also used for stability control. In the second case, collision-free path and roll stability limits are defined for path tracking. Second, in the lower motion-tracking level, to handle model uncertainties and external disturbances, a learning-based model predictive control using Gaussian process regression is designed based on the revised stability envelopes and boundaries. Finally, the effectiveness and superiority of the proposed control framework are demonstrated in various emergency scenarios. The results indicate that the control scheme effectively maintains the stability of the vehicle and improves transportation efficiency while avoiding obstacles.

The modular chassis architecture of distributed drive electric vehicles (DDEVs) provides flexibility for integrating more electrical control units. To address the challenge of enhancing longitudinal dynamics while guaranteeing ride comfort, especially under frequent urban acceleration and deceleration conditions, this paper proposes a multi-agent system (MAS)-based framework to integrate the torque vectoring system (TVS) and the active suspension system (ASS), aiming to achieve better vehicle dynamics performance. First, a half-vehicle dynamics model is constructed to describe the coupling between longitudinal and vertical motions. The polytope technique is employed to address tire nonlinearity and time-varying system states. Then, cooperative control between the TVS and ASS is developed using the MAS system, where interaction behavior is modeled based on distributed model predictive control (DMPC) optimization results, and game theory is applied to find the optimal solution. This design effectively addresses the need for modularity and scalability in integrated chassis control systems. Furthermore, terminal constraints are introduced to ensure system stability performance. Finally, the experimental tests are performed to verify the performance of the proposed MAS framework. The results demonstrate the effectiveness in enhancing vehicle longitudinal driving performance while ensuring driving comfort. Note to Practitioners-Distributed drive electric chassis, with its advantages of rapid response and precise control, has become a critical platform for autonomous vehicles. The gradual advancement of electronic control units has driven original equipment manufacturers to integrate these electronic control units to enhance chassis performance. Many leading manufacturers, including Volvo, Mercedes-Benz, BMW, and Toyota, have been at the forefront of incorporating advanced ECUs into their vehicles, such as those used for electric power steering and automatic emergency braking. Considering that frequent acceleration and deceleration in urban driving conditions are more likely to affect the vehicle's vertical motion, this work aims to integrate torque vectoring and active suspension systems to improve the vertical motion of distributed drive electric vehicles while ensuring longitudinal driving performance. However, conventional centralized control methods struggle to meet the plug-and-play requirements for integrating electronic control units. Therefore, this paper proposes a MAS-based distributed framework to achieve functional scalability and modular design. This technology is expected to become a new paradigm for the integration of electronic control units.

Shared control provides a human-centered development direction for intelligent driving. However, existing shared control methodologies address the risk factors related to both the driver and the traffic environment inadequately. To this end, a shared steering control strategy is proposed based on the driver-vehicle-road (DVR) system risk assessment result. Firstly, the driver’s steering behavior is described through a two-point preview driver model. The key parameters are identified using real driving data. Meanwhile, the deep deterministic policy gradient (DDPG) algorithm is applied to train a reinforcement learning (RL) agent considering tracking accuracy, steering smoothness and vehicle stability as the autonomous driving controller. Afterwards, three time-varying risk factors are designed to evaluate the DVR system risk level, which represent driver risk, road risk and lane departure risk, respectively. Based on the system risk level, the control authority is initially calculated by a fuzzy inference method. Then, considering the smoothness of authority transition, a model prediction control (MPC) method is applied to optimize the initial authority level in real-time. Finally, simulation and the driver-in-the-loop (DIL) experiments are performed to validate the proposed strategy. The results demonstrate that the proposed shared control strategy could reduce driving burden and demonstrates distinct superiority in terms of human-machine collaboration, driving comfort and personalized support.

Precise clamping force regulation of the electromechanical brake system (EMB) is essential for enhancing vehicle braking performance and driving safety. However, the system’s inherent uncertainties and time-varying nonlinear characteristics across different operating modes pose significant challenges to achieving accurate and stable clamping force control. To address these problems, this paper proposes a robust hybrid control strategy for precise clamping force regulation of the EMB under full-range operating conditions. First, the system structure and control framework of the EMB are analyzed, and the equivalent simplified model of each component is established for controller design. Subsequently, a novel robust hybrid controller is introduced to enable rapid and precise control of both brake clearance and clamping force. Specifically, an improved sliding mode control, a nonlinear disturbance observer, and the proportional-integral methods are integrated to mitigate the effects of friction, damping, and uncertainty, facilitating rapid elimination and precise restoration of brake clearance. Moreover, the model predictive control method combined with a position-speed double-loop structure is employed to address nonlinearities and enable precise clamping force tracking and brake clearance management. Finally, simulations and hardware-in-the-loop (HIL) tests under representative conditions are conducted to validate the proposed robust hybrid control method. The results demonstrate that the proposed method achieves satisfactory control accuracy and response speed, with 16.7% improvement in response speed and 57.1% reduction in mean steady-state tracking error compared to baseline methods, ensuring enhanced tracking precision and dynamic adaptability under uncertainties and nonlinearities.

Brake-by-wire systems can shorten braking distance, improve regenerative efficiency, and enhance safety and comfort. Braking comfort reflects perceived braking intensity and smoothness and varies considerably across drivers. However, the fixed brake servo characteristics (BSC) cannot adapt to diverse driver preferences, potentially degrading safety and comfort. To address these issues, this paper proposes a driver behavior-aware personalized control framework for electrified vehicles with decoupled brake systems (DBS). First, the driver-in-the-loop platform is built to collect data from 149 drivers under typical car-following braking conditions, with driving styles identified by the unsupervised clustering algorithm. Then, based on the feature samples obtained from real-vehicle experiments, Vacuum Booster’s BSC is modeled by the radial basis function neural network, and driver-type adaptive BSCs of DBS at different braking aggressiveness are designed. To ensure robust pressure tracking under personalized BSCs, the nonlinear cascade controller is designed, leveraging equivalent simplified models of DBS to enhance both response speed and control accuracy. Experimental results demonstrate that the proposed strategy accommodates three driver types with distinct BSC preferences, achieving mean satisfaction scores above 7.91 and satisfactory dynamic control performance, with 43% higher accuracy and over 32% lower response delay than baseline methods.

Human–machine shared control is a critical transitional paradigm toward safe and trustworthy vehicle automation, which fundamentally depends on accurate intention understanding and adaptive authority allocation. To this end, this paper proposes an intention-aware human–machine shared control framework for integrated lateral–longitudinal path tracking. Firstly, a liquid neural network (LNN)-based driver intention predictor is developed to jointly capture dynamically coupled steering and acceleration intentions, where continuous-time, input-modulated neuron dynamics enable adaptive representation of nonlinear and time-varying driving behavior. Then, an intention-aware reinforcement learning-based authority allocation strategy is designed to explicitly regulate lateral and longitudinal control authority in a closed-loop manner. Finally, the proposed framework is validated through driver-in-the-loop experiments under representative driving scenarios. The experimental results show that the proposed shared control strategy significantly improves the human–machine cooperation quality, and achieves a balanced performance among tracking accuracy, driving efficiency, ride comfort, and personalization across drivers with different skills.

In the mixed driving environment, autonomous vehicles (AVs) in the adaptive cruise control state often face bullying behaviors such as unreasonably cut-in maneuvers of human-driven vehicles (HVs), which affects the normal driving of AVs. To enhance the ability of AVs to resist bullying and realize the safe, efficient and comfortable driving, this paper proposes a longitudinal speed planning framework based on game theory. Firstly, a non-cooperative game framework between AVs and HVs and a partial cooperative game framework between ego vehicle and other AVs are constructed. Secondly, considering the longitudinal and lateral manipulation characteristics and aggressiveness of human drivers, a more realistic HVs lane-changing game strategy is established based on location advantages, comfort and traffic efficiency factors. Thirdly, according to the factors of driving safety, ride comfort and traffic efficiency of drivers and passengers, the longitudinal speed planning strategy of AVs based on game theory is designed. Combined with the method of sampling and optimization, the longitudinal speed of ego vehicle and other AVs is solved in real time. Finally, human-in-the-loop experiments verify the effectiveness of the proposed game framework in the scenarios of HVs with different aggressiveness.

Exploding increment of electric control units in the new-generation electric vehicles could be poised to diminish the system's reliability. To this end, this article presents a residual-evaluation based event-triggered mechanism (ETM) fault tolerant control framework to ensure the controller's performance as well as address the potential faults in steer-by-wire (SbW) system. First, compared with conventional ETM, a residual evaluation function is formulated as a convex solution using linear matrix Inequalities, based on which the optimal tradeoff could be achieved between robustness of the uncertainty disturbance and the sensitivity to faulty signal. Then, the threshold of residual evaluation is determined by considering transmission errors, uncertain disturbances, and fault signals. Taking into account the faults in the SbW system, the stochastic model predive controller is thus developed to obtain additional direct yaw moment, as well as ensuring the vehicle lateral stability and path tracking performance. Furthermore, chance constraints will be incorporated to guarantee that the lateral stability index is confined within the safety boundary of phase plane. Finally, the effectiveness of the proposed architecture is verified through hardware-in-loop experiments. The results illustrate that our proposed controller possesses better path tracking and lateral stability performance while controller area network resources could also be reduced.

Trajectory prediction is a critical component of autonomous driving control systems, as its accuracy and adaptability directly determine the safety and decision reliability of vehicles operating in mixed traffic environments. However, heterogeneous traffic participants in mixed traffic environments display pronounced behavioral differences. Existing models often have limited capability to accurately capture their true interaction dynamics, thereby restricting long-term prediction performance. This paper proposes a trajectory prediction method that accounts for the asymmetric interaction behaviors among heterogeneous traffic participants. Specifically, the interaction intensity between heterogeneous agents is modeled as the combination of interaction effort and interaction cost during the conflict resolution process, thereby reflecting the asymmetric social attributes among traffic participants. Then, a self-attention mechanism is employed to capture the temporal interaction features between the target vehicle and surrounding agents, while a graph Transformer is utilized to extract spatial interaction features among all participants. Furthermore, an auxiliary interaction intensity prediction task is innovatively incorporated to enhance the feature extraction process. Experimental results on the InD and SinD datasets demonstrate that the proposed asymmetric interaction-aware trajectory prediction (AIA-TP) model achieves superior long-term prediction performance compared to baselines and exhibits better adaptability in interactive scenarios.

The integration of motion planning and control within a unified architecture based on reachability theory provides an effective approach to extending the autonomous motion capability boundaries of distributed drive electric vehicles (DDEVs). This architecture is further enhanced by embedding forward set propagation methods into the dynamics analysis of DDEVs. The analytical expression of forward reachable sets (FRSs), the real-time processing of reachability constraints, and the design of integrated planning-control frameworks remain challenging. To address these challenges, this paper proposes a reachability-constrained motion planning and control integration framework (RC-MPCI) for DDEVs. The proposed RC-MPCI features strong interpretability, rigorous safety guarantees, and computational efficiency. First, a maneuver-oriented vehicle motion model is established to construct the closed-loop system dynamics. Subsequently, an FRS computation method based on sum-of-squares programming (SOSP) is proposed. It formulates an analytical expression of dynamic reachable boundaries that accounts for multi-actuator coordination, and tracking error models are introduced to ensure the reachability of the closed- loop system. Then, a constraint optimization strategy based on collision-free tunnels (CFTs) is designed, within which online motion planning and control methods are developed under a receding horizon optimization framework. Finally, the effectiveness and robustness of the proposed RC-MPCI are confirmed through virtual simulations and hardware-in-the-loop (HIL) tests. This framework, based on reachability theory, provides both theoretical and technical foundations that enable safe and efficient autonomous motion of DDEVs in highly dynamic environments.